Abstract:

Precipitation analysis is embedded in a range of important hydrological studies for hydraulic works construction and maintenance. However, flaws and limitations in records are obstacles often encountered by researchers. One feasible solution for overcoming these obstacles is to generate synthetic series. The main objective of this work is to structure and validate a model for generating synthetic rainfall series at a daily scale. A parametric model has been constructed, where the occurrences are determined by a stochastic Markov process and the cumulative rainfall quantities are computed using a mixed exponential probability distribution. Since no previous studies using the proposed probability distribution in La Plata Basin were found in the literature, several significance tests and relevant criteria were applied, in order to verify the model accuracy. The approach was studied in 11 rainfall stations inside Parana and Uruguay rivers basins, located in Brazilian South and Southeast regions, obtaining good results. Additional analyses of the model performance related to extreme events and droughts are also present.